Search

International audience; Alcoholic fermentation in oenological conditions is a biological process carried out under significant physiological constraints: deficiency of nitrogen and others nutriments (vitamins, lipids …) and different stresses (pH and osmotic). In literature, few models were proposed to describe oenological fermentations. They are focused on initial conditions and did not integrate nitrogen addition during the fermentation process which is a widespread practice. In this work, two dynamic models of oenological fermentation are proposed to predict the effects of nitrogen addition at two different timings: at the beginning of the process and during the fermentation experiment. They were validated and compared against existing models showing an accurately fit to experimental data for CO2 release and CO2 production rate.

Freezing is widely used for bacterial cell preservation. However, resistance to freezing can greatly vary depending on bacterial species or growth conditions. Our study aims at identifying cellular markers of cryoresistance based on the comparison of three lactic acid bacteria (LAB) exhibiting different tolerance to freezing: Carnobacterium maltaromaticum CNCM I-3298, Lactobacillus delbrueckii subsp. bulgaricus ATCC 11842, and Lactobacillus delbrueckii subsp. bulgaricus CFL1. A thorough characterization of their cytoplasmic membrane properties was carried out by measuring their fatty acid composition, membrane fluidity, and lipid phase transition upon cooling from 50 to -50 °C. Vitrification temperatures of the intra- and extra-cellular compartments were also quantified by differential scanning calorimetry. Additionally, the cell biochemical characterization was carried out using a recently developed Fourier transform infrared (FTIR) micro-spectroscopic approach allowing the analysis of live bacteria in an aqueous environment. The multivariate analysis of the FTIR spectra of fresh and thawed cells enabled the discrimination of the three bacteria according to their lipid, protein, and cell wall peptidoglycan components. It also revealed freezing-induced modifications of these three cellular components and an increase in bacteria heterogeneity for the two strains of L. bulgaricus, the freeze-sensitive bacteria. No cellular damage was observed for C. maltaromaticum, the freeze-resistant bacteria. Comparison of the results obtained from the different analytical methods confirmed previously reported cryoresistance markers and suggested new ones, such as changes in the absorbance of specific infrared spectral bands. FTIR microspectroscopy could be used as a rapid and non-invasive technique to evaluate the freeze-sensitivity of LAB.

Carnobacterium maltaromaticum is a species of lactic acid bacteria found in dairy, meat, and fish, with technological properties useful in food biopreservation and flavor development. In more recent years, it has also proven to be a key element of biological time–temperature integrators for tracking temperature variations experienced by perishable foods along the cold-chain. A dynamic model for the growth of C. maltaromaticum CNCM I-3298 and production of four metabolites (formic acid, acetic acid, lactic acid, and ethanol) from trehalose in batch culture was developed using the reaction scheme formalism. The dependence of the specific growth and production rates as well as the product inhibition parameters on the operating conditions were described by the response surface method. The parameters of the model were calibrated from eight experiments, covering a broad spectrum of culture conditions (temperatures between 20 and 37 °C, pH between 6.0 and 9.5). The model was validated against another set of eight independent experiments performed under different conditions selected in the same range. The model correctly predicted the growth kinetics of C. maltaromaticum CNCM I-3298 as well as the dynamics of the carbon source conversion, with a mean relative error of 10% for biomass and 14% for trehalose and the metabolites. The paper illustrates that the proposed model is a valuable tool for optimizing the culture of C. maltaromaticum CNCM I-3298 by determining operating conditions that favor the production of biomass or selected metabolites. Model-based optimization may thus reduce the number of experiments and substantially speed up the process development, with potential applications in food technology for producing starters and improving the yield and productivity of the fermentation of sugars into metabolites of industrial interest.

International audience; A conceptual model was developed to relate oral processing parameters and aroma release of cooked white rice. The conceptual model indicates that aroma release is dependent on the increase of particle surface area, the dilution effect of saliva and the diffusion of aroma from food residues that can be trapped in the buccal-pouches in the mouth. The model was validated against in vivo retro-nasal aroma release data during the consumption of rice flavoured with two aroma compounds (2-nonanone and ethyl propanoate) by five panellists. The oral processing behaviour of each subject was characterised at four different stages during oral processing by measuring bolus particle size, saliva content and the amount of residue particles that could be washed from the mouth after bolus expectoration. The results showed that aroma release for all subjects were dependent on the particle breakdown pathways used in oral processing. Subjects who reduced the rice to smaller particle sizes, higher pasted fraction and higher bolus residues had higher aroma release as expected from the conceptual model. Accounting for the physiological variables of subjects, the physicochemical parameters of aroma compounds and using a larger number of subjects in future studies will improve the model reliability.

International audience; Specific devices that combine 96-well plates and high-throughput vials were recently proposed to improve the efficiency of formulation screening. Such devices make it possible to increase the number of formulations tested while reducing the amount of active ingredients needed. The geometry of the product container influences the heat and mass transfer during freeze-drying, impacting product temperature (T_{p}) and therefore affecting the final product quality. Our study aimed to develop a tool to identify the operating conditions resulting in the same Tp when using high-throughput vials inside well plates and serum vials. Heat transfer coefficients between the shelf and the high-throughput vials (KV) were measured using the gravimetric method at chamber pressures ranging from 4 to 65 Pa for a batch of 576 vials located at the center of the well plates. KV distributions were used to predict TP distributions during primary drying of a 5% sucrose solution. Tp values were in average 8 °C higher using high-throughput vials instead of serum vials at chamber pressures lower than 12 Pa. This study provides a graphical solution for the management of process scale-up and scale-down between both types of product containers depending on their respective KV and product resistance to mass transfer.

BACKGROUND: Bioconversion broths are complex media with microorganisms that convert substrates into products in the presence of salts and nitrogen sources and that may release biomolecules. This paper deals with the impact of bioconversion broth components on the membrane‐based reactive extraction of 3‐hydroxypropionic acid (3‐HP) by tri‐n‐octylamine (TOA) in n‐decanol in preparation for an in situ product recovery. The focus here is on the influence of 3‐HP concentration (0.5–10 g L⁻¹), initial pH of the solution (3–5) and presence of proteins and salts on the extraction yield and kinetics. RESULTS: It was found that reducing the initial acid concentration caused an acceleration of the extraction kinetics because of a higher extent of complexation with TOA. pH effects were dramatic, as enhancing the pH from 3 to 5 decreased the extraction yield from 74% to only 5% due to acid dissociation. Proteins were shown to have negligible impact on the extraction yield and kinetics, probably because of their negligible mass transfer resistance at the liquid–liquid interface compared with the membrane. Conversely, the presence of salts (potassium chloride and potassium biphosphate) was highly detrimental. The decrease in extraction yield was shown to be due to an anion exchange between the carboxylate anion of 3‐HP in the organic phase and the anion of the salt in the aqueous phase. Chloride ions had more impact than biphosphate ions. CONCLUSION: These results give valuable information for the implementation of a membrane‐based reactive extraction as an in situ product recovery process, suggesting recommendations for bioconversion tuning. © 2018 Society of Chemical Industry

International audience; In the freeze-drying process, vials located at the border of the shelf usually present higher heat flow rates which in turn result in higher product temperatures than central vials. This phenomenon, named edge vial effect, can result in product quality variability within the same batch of vials and between batches at different scales. Our objective was to investigate the effect of various freeze-dryer design features on the heat transfer variability. A 3D mathematical model previously developed in COMSOL Multiphysics and experimentally validated was used to simulate heat transfer of a set of vials located at the edge and in the centre of the shelf. The design features considered were the loading configurations of the vials, the thermal characteristics of the rail, the walls and the shelves and some relevant dimensions of the drying chamber geometry. The presence of the rail in the loading configuration and the value of the shelf emissivity strongly impacted on the heat flow rates received by the vials. Conversely, the heat transfer was not significantly influenced by modifications of the thermal conductivity of the rail, the emissivity of the walls and by the geometry of the drying chamber. The developed model revealed to be a powerful tool to predict the heat transfer variability between edge and central vials for the cycle development and scale-up and to compare various freeze-dryer design features.

Vial design features can play a significant role in heat transfer between the shelf and the product and, consequently, in the final quality of the freeze-dried product. Our objective was to investigate the impact of the variability of some geometrical dimensions of a set of tubing vials commonly used for pharmaceuticals production on the distribution of the vial heat transfer coefficients (Kv) and its potential consequence on product temperature. Sublimation tests were carried out using pure water and 8 combinations of chamber pressure (4-50 Pa) and shelf temperature (-40°C and 0°C) in 2 freeze-dryers. Kv values were individually determined for 100 vials located in the center of the shelf. Vial bottom curvature depth and contact area between the vial and the shelf were carefully measured for 120 vials and these data were used to calculate Kv distribution due to variability in vial geometry. At low pressures commonly used for sensitive products (below 10 Pa), the vial-shelf contact area appeared crucial for explaining Kv heterogeneity and was found to generate, in our study, a product temperature distribution of approximately 2°C during sublimation. Our approach provides quantitative guidelines for defining vial geometry tolerance specifications and product temperature safety margins.

BACKGROUND: 3‐Hydroxypropionic acid (3‐HP) production through glycerol bioconversion by Lactobacillus reuteri suffers from low yields and productivities due to product inhibition. Reactive extraction assisted by a Hollow Fibre Membrane Contactor (HFMC) is a promising strategy for process intensification. However, the use of this integrated system is hindered by extraction phase toxicity towards the microorganism. This study describes a solvent selection strategy based on extraction performance (extraction yield and viscosity, related to mass transfer), and a low solvent toxicity, in order to find an extraction phase composition that allows continuous in stream extraction of 3‐HP. RESULTS: Inert diluent addition to a trioctylamine (TOA)‐decanol mixture decreased its toxicity and viscosity but decreased extraction yield. The linear and ramified long‐chain alcohols tested showed that increasing the number of carbon atoms decreased extraction performance as well as toxicity. Ramified alcohols showed the lowest extraction performance. Didodecylmethylamine (DDMA) gave higher extraction yield and lower solvent toxicity than TOA. Flow cytometry with dual staining for cell membrane integrity and enzymatic activity proved to give concordant and complementary information with cell bioconversion ability, being an adequate and quick method for solvent toxicity assessment. The selected organic phase consisted of 20% DDMA, 47% dodecanol and 33% dodecane by volume, and can be used for in‐stream extraction of 3‐HP produced by L. reuteri. CONCLUSIONS: The integrated selection criteria proposed in this study – extraction yield, solvent viscosity and toxicity – provide key information for choosing an organic phase with the best trade‐off between extraction performance and biocompatibility. © 2019 Society of Chemical Industry

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement N° 777657; Freezing is the most used bacterial cell preservation technique yet still a process that can be damaging and lead to cell death. Cryosensitivity greatly varies depending on considered species or strains. Fourier Transform Infra-Red (FTIR) spectroscopy is a powerful technique allowing biochemical characterization of major cellular components (lipids, proteins, polysaccharides). However, the exploitation of bacterial spectral features under native aqueous environments is challenging due to the strong absorption of water in the mid-IR region. The identification of cryoresistance markers has thus so far been mainly done using FTIR spectra of dried cells. In this study, three lactic acid bacteria strains were selected for their contrasting cryo-sensitivities. The objective was to combine two complementary approaches: a dynamic approach that measures evolution of peak positions in the lipid region of IR spectra as a function of temperature, and a novel approach using a FTIR microscope enabling higher spatial resolution of cells (~103 to 104 cells) in an aqueous environment. The most cryoresistant strain displayed a strikingly different membrane lipid phase transition compared to the other two: although phase transition happened very abruptly and at a low temperature (-14 °C), in gel phase, lipid membranes of the resistant strain maintained a higher degree of disorganization. These observations could be attributed to high unsaturated fatty acid content (>66% of C18:1). In addition to confirming the lipid membrane’s role in cryoresistance, this study has brought to light the potential role of other cellular components by removing the spectral contribution of water in spectra obtained with the FTIR microscope. This was achieved using a specially developed Matlab program, providing access to a large spectral region encompassing both the protein and carbohydrate regions (1800–975 cm-1). Good discrimination and visualization of population heterogeneity was observed, indicating that markers outside the lipid region must also contribute to cryoresistance

International audience; Reactive extraction is a well-known method for the recovery of organic acids from dilute streams. The extraction through complex formation with amines is followed by its back-extraction. Their affinity for amine extractants makes their back-extraction a limiting step. The case of 3-hydroxypropionic acid is addressed here. Three methods were compared: salts additions, diluent swing and temperature swing. Mineral bases led to complete 3-HP recovery but as a salt. Diluent swing was tested using n-hexane as the anti-solvent. Results showed that increasing the n-hexane content led to a great increase of 3-HP recovery. However, high recovery yields are associated with high dilutions. The temperature was varied from 4 to 140 °C and it led to a high decrease of the acid-amine complex formation. Our thermochemical study showed an apparent heat of reaction of-25.3 kJ/mol and, at 140 °C, 78% of 3-HP can be recovered in a single step.

[EN] Freeze-drying is a dehydration method suitable for the stabilization of heat-labile pharmaceutical products, such as vaccines. Due to the vial-to-vial variability of heat and mass transfer during the process, the value of the critical process parameters (e.g., product temperature, sublimation rate) may be different between vials and batches often present significant product quality heterogeneity. The aim of this work was the development of a dynamic, multi-vial mathematical model making it possible to predict risk of failure of the process, defined as the percentage of vials potentially rejected by quality inspection. This tool could assist the design of freeze-drying cycle., This work was sponsored by GlaxoSmithKline Biologicals SA which was involved in all stages of the study conduct and analysis

International audience; This study deals with the production of isoamyl acetate (IAA) by fermentation of sugar cane molasses with the strain Pichia fermentans ITD00165, using L-leucine as precursor. A mathematical model that describes the experimental data from fermentation was developed for its use as a tool for further process optimization. The fermentation system was constantly aerated and coupled to liquid–liquid in situ extraction with decane as the recovery solvent. Thus, the model integrates the biological production of IAA, its partition coefficient in the two liquid phase system and the stripping effect of aeration. A productivity of 26 mg L−1 h−1 was obtained with addition of 4 g L−1 of L-leucine at 12 h of fermentation. The use of the model for process optimization was explored. According to it, the maximum theoretical productivity that can be obtained is 63 mg L−1 h−1. The model was used to determine that 1.6 g L−1 is the minimum concentration of L-leucine that can be added without significantly reducing IAA production. Also, it makes possible to propose an adequate decane/culture medium ratio, to have a desired final concentration and amount of recovered IAA. This value can be adjusted based on the needs of further purification steps and is useful to define a global economic optimum of the process.

International audience; During the primary drying step of the freeze-drying process, mass transfer resistance strongly affects the product temperature, and consequently the final product quality. The main objective of this study was to evaluate the variability of the mass transfer resistance resulting from the dried product layer (R_p) in a manufacturing batch of vials, and its potential effect on the product temperature, from data obtained in a pilot scale freeze-dryer. Sublimation experiments were run at -25 °C and 10 Pa using two different freezing protocols: with spontaneous or controlled ice nucleation. Five repetitions of each condition were performed. Pressure rise test (PRT) and gravimetric methods were applied as complementary approaches to estimate R_p. The PRT allowed to assess variability of the evolution of R_p with the dried layer thickness between different experiments whereas the gravimetric method informed about R_p variability at a fixed time within the vial batch. A product temperature variability of approximately ± 4.4 °C was defined for a product dried layer thickness of 5 mm. The present approach can be used to estimate the risk of failure of the process due to mass transfer variability when designing freeze-drying cycle.

Can the desorption kinetics explain the residual moisture content heterogeneity observed in pharmaceuticals freeze-drying process?. 6. European Drying Conference: EuroDrying 2017

International audience; In pharmaceutical freeze-drying, the position of the product container (vial) on the shelf of the equipmentconstitutes a major issue for the final product quality. Vials located at the shelf edges exhibit higherproduct temperature than vials located in the centre, which in turn often results in collapsed product. Aphysics-based model was developed to represent heat transfer phenomena and to study their variationwith the distance from the periphery of the shelf. Radiation, conduction between solids, and conductionthrough low-pressure water vapour were considered. The modelling software package COMSOLMultiphysics was employed in representing these phenomena for a set of five vials located at the borderof the shelf, close to the metallic guardrail. Model predictions of heat fluxes were validated against experimentalmeasurements conducted over a broad range of shelf temperatures and chamber pressures representativefor pharmaceutical freeze-drying. Conduction through low-pressure water vapour appearedas the dominant mechanism explaining the additional heat transfer to border vials compared to centralones. The developed model constitutes a powerful tool for studying heterogeneity in freeze-drying whilereducing experimental costs.

A comprehensive study of equilibrium states involved in the reactive extraction of 3-hydroxypropionic acid (3-HP) by tri- n -octylamine (TOA) in n -decanol is described. Complexation phenomena between 3-HP and TOA were revealed by infrared spectroscopy (FT-IR). This study demonstrated that the main extraction mechanism is the formation of an ion pair involving the dissociated form of 3-HP. Chemical models were then formulated, taking the formation of the ion pair between 3-HP and TOA and the influence of the amount of solvent in the organic phase into account. Two models are proposed: (1) a stoichiometric model where the solvent is seen as a reagent that is involved in the complexation as a synergistic extractant, but with competition for solvation via H-bond interactions with TOA; (2) a model where the solvent is seen as a phase modifier that improves the physicochemical properties of the extractant and that changes the complexation equilibrium. Both models describe the extraction yields in the investigated ranges, with Model (2) being particularly predictive since it provides a mean absolute prediction error of less than 2% in yield units. According to this latter model, the complexation equilibrium coefficient increases with the increase of the n -decanol proportion in the organic phase. Model (2) shows that solvation effects are better described as non-stoichiometric interactions. Therefore, the two mechanistic models accurately represent extraction yields over a wide range of initial conditions and are potentially transposable to other reactive extraction systems. Such knowledge is essential for further implementation of downstream processes.

Eco-design of food and bio-processes: Example of production of stabilized lactic acid bacteria. Lecture at WUR

To evaluate the effects of temperature and relative humidity (RH) on microbial and biochemical ripening kinetics, Camembert-type cheeses were prepared from pasteurized milk seeded with Kluyveromyces marxianus, Geotrichum candidum, Penicillium camemberti, and Brevibacterium aurantiacum. Microorganism growth and biochemical changes were studied under different ripening temperatures (8, 12, and 16 degrees C) and RH (88, 92, and 98%). The central point runs (12 degrees C, 92% RH) were both reproducible and repeatable, and for each microbial and biochemical parameter, 2 kinetic descriptors were defined. Temperature had significant effects on the growth of both K. marxianus and G. candidum, whereas RH did not affect it. Regardless of the temperature, at 98% RH the specific growth rate of P. camemberti spores was significantly higher [between 2 (8 degrees C) and 106 times (16 degrees C) higher]. However, at 16 degrees C, the appearance of the rind was no longer suitable because mycelia were damaged. Brevibacterium aurantiacum growth depended on both temperature and RH. At 8 degrees C under 88% RH, its growth was restricted (1.3 x 10(7) cfu/g), whereas at 16 degrees C and 98% RH, its growth was favored, reaching 7.9 x 10(9) cfu/g, but the rind had a dark brown color after d 20. Temperature had a significant effect on carbon substrate consumption rates in the core as well as in the rind. In the rind, when temperature was 16 degrees C rather than 8 degrees C, the lactate consumption rate was approximately 2.9 times higher under 88% RH. Whatever the RH, temperature significantly affected the increase in rind pH (from 4.6 to 7.7 +/- 0.2). At 8 degrees C, an increase in rind pH was observed between d 6 and 9, whereas at 16 degrees C, it was between d 2 and 3. Temperature and RH affected the increasing rate of the underrind thickness: at 16 degrees C, half of the cheese thickness appeared ripened on d 14 (wrapping day). However, at 98% RH, the underrind was runny. In conclusion, some descriptors, such as yeast growth and the pH in the rind, depended solely on temperature. However, our findings highlight the fact that the interactions between temperature and RH played a role in P. camemberti sporulation, B. aurantiacum growth, carbon substrate consumption rates, and the thickening of the cheese underrind. Moreover, the best ripening conditions to achieve an optimum between microorganism growth and biochemical kinetics were 13 degrees C and 94% RH.

A multidisciplinary approach combining physiology and physical chemistry and associating experimental measurements with in silico modelling was applied to explain the release of aroma compounds during food consumption. Experimental release kinetics obtained by inhaling gaseous samples through controlled protocols highlighted different release behaviours, depending on aroma compound properties. The associated mathematical model described mass transfer mechanisms between the different compartments of the naso-oro-pharyngeal cavities and included both physicochemical and physiological parameters. One of the main developments was notably to consider the possible retention of aroma compounds by wetted mucosa. Model sensitivity analysis confirmed the key role of interaction between aroma compounds and mucosa (air/mucosa partition coefficient) and of individual breath parameters (current breath volume and respiratory frequency) on the persistence of aroma compound in exhaled air. These achievements show that the association of an experimental approach and mechanistic modelling constitutes a powerful tool to improve the understanding of aroma release and persistence.

The aim of this study was to gain knowledge about the role of triacylglycerol (TAG) composition in fatty acids (FA) of o/w emulsions on both the pancreatic lipolysis kinetics and the bioaccessibility of released products (i.e. contained within the bile salt micellar phase). A mathematical model was developed and its predictions were compared to a set of experimental data obtained during an in vitro digestion of a whey protein stabilized emulsion. Modeling results show that FA residues of TAG were hydrolyzed at specific rates, inducing different bioaccessibility kinetics. The estimated lipolysis rate constants of the studied FA (C8:0, C10:0 >> C18:1 n-9 >> C12:0 > C14:0 > C16:0 approximate to C16:1 n-7 > C22:6 n-3) were in close agreement with the available literature on the substrate specificity of pancreatic lipase. Results also suggest that lipolysis products are very rapidly solubilized in the bile salt mixed micelles with no fractionation according to the FA carbon chain.

Submicron oil-in-water emulsions stabilized with whey proteins have been reported to be relatively resistant and highly sensitive to coalescence during the gastric and intestinal phases of in vitro digestion experiments, respectively. The aim of this study was to assess the impact of droplet coalescence on the intestinal lipolysis kinetics of such emulsion, and to develop a mathematical model able to predict the experimental observations. A submicron whey protein stabilized emulsion made of a mixture of medium-chain (MCT) and long-chain triacylglycerols (LCT) was prepared and submitted to gastrointestinal in vitro digestion. Triacylglycerol concentrations and droplet size distributions were measured before and after the gastric phase and during the intestinal phase using HPLC and laser granulometry, respectively. MCT were fully digested within 15 min of intestinal digestion, whereas LCT were still detected after 5 h. Moreover, the intestinal lipolysis of LCT showed a two-stage behavior with an initial fast rate that markedly slowed down after about 30 min, a time at which a sudden rise in the droplet sizes, attributed to coalescence, was also observed. A mathematical model based on the experimentally measured droplet sizes and assuming a rate of lipolysis proportional to the interfacial area was developed and successfully used to reproduce the observed kinetics. Our results support the idea that droplet coalescence during the intestinal phase was the main reason for the marked slowdown of the kinetics of lipid digestion, hence suggesting that other possible mechanisms, such as an inhibition of the lipolysis reaction, could only explain secondary order effects.

International audience; A novel dynamic model was developed for predicting the synthesis kinetics of the principal aroma compounds produced by yeasts during winemaking fermentations: isobutanol, isoamyl alcohol, isoamyl acetate, ethyl hexanoate and ethyl octanoate. The parameters of the model were identified from nine fermentations performed at temperatures between 18 and 30 degrees C and with different initial nitrogen contents, in the range of 70 to 410 mgN/L. The model was validated in six independent experiments with conditions in the same range. Predictions were accurate for these volatile compounds: the mean difference between experimental and estimated values for fermentative aroma synthesis throughout the process was below 10%, for both the fermentations used to build the model and those used for validation. This model is the first to simulate the production kinetics of fermentative aromas and provides new insight into the synthesis of these volatile compounds. It will facilitate the development of innovative strategies for controlling the production of those aromas in winemaking, through management of the principal control factors: assimilable nitrogen content and temperature during the alcoholic fermentation.

Ce numéro d'Innovations Agronomiques comprend les articles correspondant aux présentations du colloque « Comprendre et utiliser la structure des aliments pour améliorer leurs qualités nutritionnelles et sensorielles » qui s'est tenu à Rennes le 13 mai 2014.; The design of food, as smart vectors of target molecules responsible for both sensory and nutritional properties, is a major challenge for the food industry. The aim is to provide good and healthy foods,based on the specific needs of populations (children, athletes, elderly or overweight people). The digestive tract, from the mouth to the intestine, can be considered as a multistep reactor, compact and perfectly adaptable. During this process, i) the structure and therefore the physical characteristics of the bolus, and ii) the physiological characteristics of the individual impact the kinetics of digestion and influence the physiological control related to health and well-being. Mathematical modeling of transfers and reactions in the digestive tract is an original and powerful approach. It allows the decoupling of mechanisms linked to the structure and composition of the food on the one hand and to the physiology of individual on the other hand. It thus provides a basis for building tools to help in the food designing. This approach, coupling in vitro and in vivo experiments and in silico developments, is illustrated through three examples based on dairy products of different composition and structures (model gels, cheeses, yogurts) in order to predict the flavour perception (salt and aroma) and the digestion of proteins.; La conception d'aliments, en tant que vecteurs intelligents de molécules cibles responsables aussi bien de propriétés sensorielles que nutritionnelles correspond à un enjeu majeur pour l'industrie alimentaire. Il s'agit de proposer des aliments avec des teneurs réduite en sel, sucre ou matière grasse, ou bien enrichis en fibres ou protéines, et ceci en intégrant les besoins spécifiques des populations (enfants, sportifs, personnes agées ou en surpoids ...). Le tube digestif, de la bouche jusqu'à l'intestin, peut être considéré comme un formidable réacteur multi étagé, compact et parfaitement adaptable. Au cours de ce processus, la structure et donc les caractéristiques physiques du bol alimentaire, ainsi que les caractéristiques de l'individu qui consomme l'aliment impactent les cinétiques de digestion et influencent les régulations physiologiques associées à la santé et au bien-être de l'Homme. La modélisation mathématique des réactions et transferts dans le tube digestif constitue une approche originale et puissante, car elle permet de découpler les phénomènes liés à la structure et la composition de l'aliment d'une part et à l'individu d'autre part. Elle constitue ainsi une base pour construire des outils d'aide à la formulation. Des approches couplant mesures in vitro et in vivo associées à des développements in silico sont illustrées au travers de trois exemples sur des produits laitiers de différentes structures et compositions (gels laitiers modèles, yaourts et fromages), et ceci afin de prédire la perception salée, la perception d'arômes et enfin la digestion de protéines.

Research Areas: Life Sciences & Biomedicine - Other Topics; Mathematical & Computational Biology; A mechanistic model predicting flavour release during oral processing of masticated foods was developed. The description of main physiological steps (product mastication and swallowing) and physical mechanisms (mass transfer, product breakdown and dissolution) occurring while eating allowed satisfactory simulation of in vivo release profiles of ethyl propanoate and 2-nonanone, measured by Atmospheric Pressure Chemical Ionization Mass Spectrometry on ten representative subjects during the consumption of four cheeses with different textures. Model sensitivity analysis showed that the main parameters affecting release intensity were the product dissolution rate in the mouth, the mass transfer coefficient in the bolus, the air-bolus contact area in the mouth and the respiratory frequency. Parameters furthermore affecting release dynamics were the mastication phase duration, the velopharynx opening and the rate of saliva incorporation into the bolus. Specific retention of 2-nonanone on mucosa was assumed to explain aroma release kinetics and confirmed when gaseous samples were consumed.

The aim of this work was to elaborate a predictive model of the mass transfer (water loss and solute gain) that occurs during dewatering and soaking by using neural network modelling. Two separate feedforward networks with one hidden layer were used (for water loss and solute gain respectively). Model validation was carried out on results obtained previously, which dealt with agar gel soaked in sucrose solution over a wide experimental range (temperature, 30-70 °C; solu tion concentration, 30-70 g sucrose/100 g solution; time 0-500 min; agar concentration, 2-8%). The best results were obtained with three hidden neurons, which made it possible to predict mass transfer, with an accuracy at least as good as the experimental error, over the whole experimental range. The technological interest of such a model is related to a rapidity in simulation compa rable to that of a traditional transfer function, a limited number of parameters and experimental data, and the fact that no preliminary assumption on the underlying mechanisms was needed.

A predictive optimal control algorithm for on-line control of a batch drying process is presented. An optimal command profile is computed off-line and is corrected on-line taking into account feedback measurements. A simple non-linear process model is used for predictions in both off-line and on-line optimization. The problem is entirely stated in time domain and includes technological constraints.

In most cases, industrial mixed flow corn dryers are controlled manually or by using very simple heuristic procedures, with a relatively poor performance. A detailed analysis, carried out using both a first principles model and a black-box approach, shows important non-linearities and gives a better insight into plant dynamics which may explain difficulties in using conventional linear control. Prior knowledge is used to derive a combined logarithmic and variable sample period co-ordinate transformation which is shown to linearize the plant exactly under reasonable simplifying assumptions. The resulting linear model has a rather special structure and better approximation capabilities than common transfer functions with many more estimated parameters. The control implications of the model are analysed. A simple PI controller is shown to be a suitable choice for an industrial application and significant improvements cannot be expected from more sophisticated designs such as LQG.

Non-linear dynamic models of corn drying and wet-milling quality degradation are obtaineddirectly from experimental data. No assumptions about the underlying mechanisms are made. Relative advantages of recurrent versus explicit-time models and technical issues are discussed. This type of model is well suited for very fast on-line simulations, for example in a predictive optimal control algorithm.

Developing smart control strategies of freeze-drying of lactic acid bacteria. 5. International Conference on Lyophilization and Freeze Drying

In vivo aroma release during solid food consumption is a complex phenomenon that depends on food structure and composition, as well as on oral processing (combination of mastication and incorporation of saliva into the food product). The objective of this study was to understand and to predict the physicochemical properties of aroma compounds through the dynamics of flavor release during in-mouth oral processing of food before bolus swallowing. Within this context, the evolution of two aroma compounds during bolus formation was explored by studying the two main properties that account for mass transfer: air/ bolus partition and mass transfer coefficients. Four types of industrial cheese products (varying in fat and firmness) flavored with ethyl propanoate and 2-nonanone were chosen. Each matrix was mixed with various amounts of artificial saliva to mimic boluses at different stages of mastication. The air/bolus partition coefficient was determined by the static phase ratio variation method (PRV), while the mass transfer coefficient was obtained by non-linear regression from dynamic headspace experiments. Results showed that there is a dilution effect on the air/bolus partition coefficient and both a dilution and a product effect (firmness) on the mass transfer coefficient of ethyl propanoate in the bolus. These results were also validated with 2-nonanone for the low-fat cheeses.

The aim of this review is twofold. Firstly, we present the state of the art in dynamic modelling and model-based design, optimisation and control of food systems. The need for nonlinear, dynamic, multi-physics and multi-scale representations of food systems is established. Current difficulties in building such models are reviewed: incomplete, piecewise available knowledge, spread out among different disciplines (physics, chemistry, biology and consumer science) and contributors (scientists, experts, process operators, process managers), scarcity, uncertainty and high cost of measured data, complexity of phenomena and intricacy of time and space scales. Secondly, we concentrate on the opportunities offered by the complex systems science to cope with the difficulties faced by food science and engineering. Newly developed techniques such as model-based viability analysis, optimisation, dynamic Bayesian networks etc. are shown to be relevant and promising for design and optimisation of foods and food processes based on consumer needs and expectations.

International audience; The temporal dominance of sensations method showed that four model dairy products had different dynamic profiles in terms of salt and texture perception. We investigated the physical origins of these differences, by studying the breakdown of these products and its impact on salt release. An experimental device was used for monitoring the kinetics of salt release from the food products into water -simulating saliva- after a standardized compression -simulating mastication independently of the inter and intra individual variability. A mechanistic model was developed to quantify product breakdown in terms of the area of contact between the product and the aqueous phase. Fat had a major influence on breakdown behaviour and the calculated contact area that could be accounted for by the microstructure of the product. These results provide insight into the possible origins of differences in sensory perceptions of foods. We also discussed the use of this mechanistic model for modeling salt release in the conditions of food consumption.

International audience

International audience

International audience; On the basis of a first mechanistic model predicting aroma release in the oropharynx during food consumption, the aim of the present work was to improve its accuracy and to use it to identify the main mechanisms responsible for in-mouth aroma release. Comparison between predicted release kinetics and the ones measured by APCI-MS in the nasal cavity of subjects eating flavoured yogurt highlighted the reasonably accurate time predictions of the relative aroma concentration in the nasal cavity and the model ability to simulate successive swallowing events as well as partial velopharyngeal closure. Parameters identified as the most influent for in-vivo aroma release were related to either physico-chemistry (equilibrium air/product partition coefficient, mass transfer coefficient in the product), physiology (the volume of the nasal cavity, breath airflow rate) or interaction between both (the amount of residual product in the pharynx, air/product contact area).

International audience

International audience

International audience

A model predicting the weight loss and the dry matter increase in the core and in the rind of a French smear soft cheese (Munster type) has been developed. Experimental data were collected from experiments carried out in an aseptic pilot scale ripening chamber under nine different combinations of temperature (8, 12 and 16 °C) and of relative humidity (85%, 93% and 99%) according to a complete experimental design. For each run 16 samples were analysed at various ripening stages. The model took into account the water loss by evaporation, the water transfer from the core to the surface and the mass loss due to the CO2 release. The cheese weight loss was more influenced by the relative humidity than by the temperature. The model was able to account for the experimental weight loss as well as the rind and the core dry matter, with accuracy acceptable for practical use.

A mechanistic model for Debaryomyces hansenii growth and substrate consumption, lactose conversion into lactate by lactic acid bacteria, as well as lactose and lactate transfer from the core toward the rind was established. The model described the first step (14 d) of the ripening of a smear soft cheese and included the effects of temperature and relative humidity of the ripening chamber on the kinetic parameters. Experimental data were collected from experiments carried out in an aseptic pilot scale ripening chamber under 9 different combinations of temperature (8, 12, and 16 degrees C) and relative humidity (85, 93, and 99%) according to a complete experimental design. The model considered the cheese as a system with 2 compartments (rind and core) and included 5 state evolution equations and 16 parameters. The model succeeded in predicting D. hansenii growth and lactose and lactate concentrations during the first step of ripening (curd deacidification) in core and rind. The nonlinear data-fitting method allowed the determination of tight confidence intervals for the model parameters. The residual standard error (RSE) between model predictions and experimental data was close to the experimental standard deviation between repeated experiments.

Aroma compound properties in food matrices, such as volatility and diffusivity, have to be determined to understand the effect of composition and structure on aroma release and perception. This work illustrates the use of mass transfer modeling to identify diffusion and partition properties of ethyl hexanoate in water and in carrageenan matrices with various degrees of structure. The comparison of results obtained with a diffusive model to those obtained with a convective model highlights the importance of considering the appropriate transfer mechanism. Modeling of the preliminary experimental steps ensures correct estimation of the conditions for the main aroma release step. The obtained values of partition and diffusion coefficients are in agreement with those found in the literature (either experimentally determined or predicted by theoretical equations) and demonstrate that the structure level of carrageenan matrices has little influence on diffusion properties of ethyl hexanoate (less than 20%).

International audience

International audience

National audience

International audience

Mass yield and operating time for a pork meat dehydration–impregnation–soaking (DIS) process were optimized using a coupled genetic algorithm/sequential quadratic programming method in order to obtain the optimal operating conditions: temperature and soaking solution concentrations. The DIS process was simulated by a neural network model. The non-linear optimization problem was constrained to ensure the main product characteristics: stability indicated by the water activity target and flavour characterized by the phenol gain target. The climatic conditions, the model validity region, the raw material costs and the operator working schedule were taken into account. Optimal solutions are discussed for three different batch configurations: single-stage processing under constant conditions, single-stage processing under varying temperature and two-stage processing under constant conditions. The most convenient operation resulted in a two-stage soaking process because of time, energy and cost savings, control convenience, product cooling anticipation and a reasonably high mass yield.